anycall-py 0.1.3

Polyglot RPC over Redis — call functions across languages

Python - AnyCall

This directory contains the Python components of the AnyCall project. Call functions across Python services using Redis, with no .proto files, exposed ports, or extra service plumbing to maintain.

Structure

• AnyCall core library in this directory
• example-consumer/ - Consumer example
• example-supplier/ - Supplier example
• Dockerfile - Docker image for the project
• pyproject.toml - uv workspace configuration

How to use

Install

# Using uv (recommended)
uv sync

# Or using pip
pip install -e .

Quick usage

• Supplier: decorate methods with @supply, register the class, and start the server.
• Consumer: create an AnyCall client and call the supplier method by name.

server = AnyCall.server(redis_uri)
server.register(SentimentAnalyzer())
server.start()

client = AnyCall.client(redis_uri)
sentiment = client.call("analyze-sentiment", request, Sentiment)

For details on how to use as a supplier or consumer, see USAGE.md.

Modules

lib

Core library implementing an RPC framework via Redis Streams. Uses the AnyCall factory class to create clients and servers:

• AnyCall Client: Synchronous interface for invoking remote methods. Serializes the request to JSON, publishes to a Redis Stream, waits for the response on a callback stream. Supports configurable timeout and optional metrics collection.

• AnyCall Server: Listener that processes requests from Redis. Maintains a thread pool (one per registered method) consuming from specific streams. Methods are discovered via the @supply decorator on registered classes.

• Configuration: Via AnycallProperties — defines parameters like Redis URI, timeouts, thread pools, etc.

example-consumer

Client application demonstrating the use of AnyCallClient. Makes 100 RPC calls to the analyze-sentiment method and displays latency statistics (min, avg, p50, p95, p99, max). Includes warmup call and metrics support.

example-supplier

Server application that registers suppliers via AnyCall.server(). The SentimentAnalyzer class contains an analyze_sentiment method decorated with @supply("analyze-sentiment"), which handles sentiment analysis. Starts listeners for each registered method and writes a health file to /tmp/anycall/health.